Explicit Memory Tracker with Coarse-to-Fine Reasoning for Conversational Machine Reading

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 935–945 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 935 Explicit Memory Tracker with Coarse-to-Fine Reasoning for Conversational Machine Reading Yifan Gao†∗, Chien-Sheng Wu‡, Shafiq Joty‡, Caiming Xiong‡, Richard Socher‡, Irwin King†, Michael R. Lyu†, and Steven C.H. Hoi‡ † The Chinese University of Hong Kong ‡ Salesforce Research †{yfgao,king,lyu}@cse.cuhk.edu.hk ‡{cswu,sjoty,cxiong,rsocher,shoi}@salesforce.com Abstract The goal of conversational machine reading is to answer user questions given a knowl- edge base text which may require asking clar- ification questions. Existing approaches are limited in their decision making due to strug- gles in extracting question-related rules and reasoning about them. In this paper, we present a new framework of conversational ma- chine reading that comprises a novel Explicit Memory Tracker (EMT) to track whether con- ditions listed in the rule text have already been satisfied to make a decision. More- over, our framework generates clarification questions by adopting a coarse-to-fine rea- soning strategy, utilizing sentence-level entail- ment scores to weight token-level distribu- tions. On the ShARC benchmark (blind, held- out) testset, EMT achieves new state-of-the- art results of 74.6% micro-averaged decision accuracy and 49.5 BLEU4. We also show that EMT is more interpretable by visualiz- ing the entailment-oriented reasoning process as the conversation flows. Code and mod- els are released at https://github.com/ Yifan-Gao/explicit_memory_tracker. 1 Introduction In conversational machine reading (CMR), ma- chines can take the initiative to ask users ques- tions that help to solve their problems, instead of jumping into a conclusion hurriedly (Saeidi et al., 2018). In this case, machines need to understand the knowledge base (KB) text, evaluate and keep track of the user scenario, ask clarification ques- tions, and then make a final decision. This inter- active behavior between users and machines has gained more attention recently because in practice users are unaware of the KB text, thus they cannot provide all the information needed in a single turn. ∗ This work was mostly done when Yifan Gao was an intern at Salesforce Research Asia, Singapore. Statutory Maternity Pay To qualify for SMP you must: * earn on average at least £113 a week * give the correct notice * give proof you’re pregnant Do I qualify for SMP? I've been old enough to get my pension. Do you earn on average at least £113 a week? Yes No ule xt er ario tial stion n 1 n 2 n 3 Yes No Irrelevant Inquire Did you give the correct notice? Decision: Yes No Irrelevant Inquire Decision: Yes No Irrelevant Inquire Decision: No ## Taking more leave than the entitlement If a worker has taken more leave than they’re entitled to, their employer must not take money from their final pay unless it’s been agreed beforehand in writing. The rules in this situation should be outlined in the employment contract, company handbook or intranet site. Can my employer take money from my final pay? I have questions regarding my employer … Did you take more …

Content truncated to 500 words